A Multi-Platform Arabic News Comment Dataset for Offensive Language Detection

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6203–6212 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6203 A Multi-Platform Arabic News Comment Dataset for Offensive Language Detection Shammur A Chowdhury, Hamdy Mubarak, Ahmed Abdelali, Soon-gyo Jung, Bernard J Jansen, Joni Salminen Qatar Computing Research Institute Hamad Bin Khalifa University, Doha, Qatar {shchowdhury, hmubarak, aabdelali, sjung, bjansen, jsalminen}@hbku.edu.qa Abstract Access to social media often enables users to engage in conversation with limited accountability. This allows a user to share their opinions and ideology, especially regarding public content, occasionally adopting offensive language. This may encourage hate crimes or cause mental harm to targeted individuals or groups. Hence, it is important to detect offensive comments in social media platforms. Typically, most studies focus on offensive commenting in one platform only, even though the problem of offensive language is observed across multiple platforms. Therefore, in this paper, we introduce and make publicly available a new dialectal Arabic news comment dataset, collected from multiple social media platforms, including Twitter, Facebook, and YouTube. We follow two-step crowd-annotator selection criteria for low-representative language annotation task in a crowdsourcing platform. Furthermore, we analyze the distinctive lexical content along with the use of emojis in offensive comments. We train and evaluate the classifiers using the annotated multi-platform dataset along with other publicly available data. Our results highlight the importance of multiple platform dataset for (a) cross-platform, (b) cross-domain, and (c) cross-dialect generalization of classifier performance. Keywords: Offensive Language, Arabic, Text Classification, Social Media Platforms, Twitter, Facebook, YouTube 1. Introduction Social media platforms provide access for users from all over the world to connect with each other, share their knowledge, and express their opinions (Intapong et al., 2017). Within these platforms, people not only voice their opinions and concerns but also look for social connection, belongingness, and assurance of being accepted by society. Thus, any negative perceived image of a group or an indi- vidual, by a small fraction of the community, can impact users’ psychological well-being (G¨ulac¸tı, 2010; Waldron, 2012), e.g., by creating propaganda and giving rise to hate crimes. Often, this type of abusive propaganda spreads through ei- ther by the content posted in the platforms or by the toxic behavior of users in the social media. Users may use offen- sive post/comments containing vulgar, pornographic and/or hateful language, to spread such verbal hostility. Hence, the increased risk and effect of such hostility using offensive language, in social media, has attracted many multidisci- plinary researchers and provoked the need of automatically detecting the offensiveness of post/comments. Various classification techniques have been studied and used to detect offensive language (Davidson et al., 2017; Mubarak and Darwish, 2019), along with related categories of hate speech (Badjatiya et al., 2017; Salminen et al., 2018), cyberbullying (Dadvar et al., 2013; Hosseinmardi et al., 2015), aggression (Kumar et al., 2018) among oth- ers. Modeling techniques such as keyword-based search, traditional machine learning to deep learning have been ex- plored. However, most of …

Content truncated to 500 words